Studying the interfacial exchange coupling within ferrite based magnetic nanoparticles prepared following to a succession of thermal decomposition synthesis

(1)

HAL Id: tel-02470768

https://tel.archives-ouvertes.fr/tel-02470768

Submitted on 7 Feb 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Studying the interfacial exchange coupling within ferrite

based magnetic nanoparticles prepared following to a

succession of thermal decomposition synthesis

Kevin Sartori

To cite this version:

Kevin Sartori. Studying the interfacial exchange coupling within ferrite based magnetic nanoparticles prepared following to a succession of thermal decomposition synthesis. Materials. Université de Strasbourg, 2019. English. �NNT : 2019STRAE029�. �tel-02470768�

(2)

UNIVERSITÉ DE STRASBOURG

ÉCOLE DOCTORALE de Physique et Chimie-Physique (ED182)

IPCMS (UMR 7504 CNRS – Unistra)

Synchrotron SOLEIL

Laboratoire Léon Brillouin (UMR 12 CEA – CNRS)

THÈSE

présentée par :

Kevin SARTORI

soutenue le : 7 novembre 2019

pour obtenir le grade de :

Docteur de l’université de Strasbourg

Discipline/ Spécialité

: Science des matériaux

Etude du couplage d’échange

interfacial au sein de nanoparticules

magnétiques à base de ferrite

préparées via une succession de

synthèses par décomposition

thermique

THÈSE dirigée par :

M. PICHON Benoît Maître de conférences, IPCMS, Université de Strasbourg

M. CHOUEIKANI Fadi Chercheur, Synchrotron SOLEIL

RAPPORTEURS :

Mme CHANEAC Corinne Professeur, INSP, Université Pierre et Marie-Curie

M. SAINCTAVIT Philippe Directeur de recherche, IMPMC, Université Pierre et Marie-Curie

AUTRES MEMBRES DU JURY :

Mme VIART Nathalie Professeur, IPCMS, Université de Strasbourg

M. CHAUDRET Bruno Directeur de recherches, LPCNO, INSA, Université Paul Sabatier

M. CHABOUSSANT Grégory Directeur de recherches, LLB, UMR12 CEA-CNRS, Gif-sur-Yvette

(3)

(4)

Illustration of a core@shell@shell nanoparticle which structure and magnetic properties were investigated by the use of a wide panel of analysis techniques.

(5)

(6)

Content

General introduction ... 15

Fe3-dO4 structure ... 18

Crystallographic structure ... 18

Magnetic structure ... 19

Size reduction to the nanoscale... 20

Magnetic domains size ... 20

Superparamagnetism ... 21

Dipolar interactions ... 23

Small iron oxide nanoparticles... 24

Exchange-bias coupling in nanoparticles ... 24

Definition and characteristics ... 24

Limitations of the exchange-bias coupling ... 27

Applications ... 29

Exchange-biased iron oxide based magnetic nanoparticles ... 29

Surface anisotropy ... 30

Phase oxidation ... 30

Synthesis of iron oxide nanoparticles ... 31

Thermal decomposition method ... 31

Seed-mediated growth synthesis by thermal decomposition ... 32

MnO, CoO and NiO wüstite phases as shells ... 33

Fe3-dO4@MnO nanoparticles ... 34

Fe3-dO4@NiO nanoparticles ... 34

Fe3-dO4@CoO nanoparticles ... 35

AFM@FiM core@shell nanoparticles... 35

FiM@FiM hard-soft magnetic exchange coupled core@shell nanoparticles ... 37

Onion-type magnetic nanoparticles ... 38

Conclusion of the introduction ... 40

Complementarity of analysis techniques ... 42

Transmission electron microscopy... 42

X-ray diffraction ... 42

Fourier transform infrared ... 43

Granulometry... 43

Small-angle X-ray scattering ... 44

(7)

Big instruments ... 44

X-ray absorption (XAS, XMCD) ... 44

Small-angle neutron scattering ... 45

Magnetometry measurements ... 45

Conclusion on the complementarity of analysis technics ... 46

(8)

CHAPTER I

Increasing the size of Fe

3-d

O

4

nanoparticles by performing a multistep

seed-mediated growth approach

Fe

3-d

O

4

(@Fe

3-d

O

4

)n

Introduction ... 57

Experimental Section ... 59

Synthesis of Fe (II) stearate precursor ... 59

Synthesis of 10 nm iron oxide cores ... 59

Addition of an iron oxide layer ... 59

Characterization techniques ... 60

Results and discussion... 62

Synthesis strategy ... 62

Fourier-transform infra-red and granulometry ... 64

High-resolution transmission electron microscopy ... 65

Mössbauer spectroscopy ... 68

SQUID magnetometry measurements ... 70

General discussion ... 73

Conclusion ... 74

(9)

Chapter II

Strong interfacial coupling through exchange interactions in soft/hard

core-shell nanoparticles as function of cationic distribution

Fe

3-d

O

4

– Co doped, Fe

3-d

O

4

@CoFe

2

O

4

, Fe

3-d

O

4

@CoO

Introduction ... 79

Experimental section... 81

Nanoparticles synthesis ... 81

Precursor synthesis ... 81

Core shell nanoparticles synthesis ... 81

SQUID magnetometry ... 91

Element specific hysteresis ... 93

Conclusion ... 97

(10)

CHAPTER III

Exchange-biased hybrid magnetic nanoparticles

Fe

3-d

O

4

@CoO@Fe

3-d

O

4

Introduction ... 101

Iron stearate precursor synthesis ... 102

Cobalt stearate precursor synthesis... 102

Iron oxide core synthesis ... 102

Fe3-dO4@CoO core@shell nanoparticles synthesis ... 102

Fe3-dO4@CoO@Fe3-dO4 core@shell@shell nanoparticles synthesis ... 103

Transmission Electron Microscopy ... 107

Fast Fourier infra-red spectroscopy ... 112

Granulometry... 114

Small-angle X-ray scattering ... 115

Polarized- small angle neutron scattering ... 136

Conclusion ... 147

(11)

CHAPTER IV

Exchange-coupled nanomagnet

Fe

3-d

O

4

@CoFe

2

O

4

@Fe

3-d

O

4

Iron stearate precursor synthesis ... 156

Cobalt stearate precursor synthesis... 156

C, CS and CSS nanoparticle synthesis ... 156

Electron microscopy ... 159

Fourier transform infra-red ... 166

Granulometry... 168

X-ray absorption... 173

Conclusion ... 185

(12)

Chapter V

AFM proximity effect of NiO on Fe

3-d

O

4

@CoO nanoparticles

Fe

3-d

O

4

@CoO@NiO

FT-IR spectroscopy ... 196

Granulometry measurements ... 197

Selective hysteresis ... 205

Summary and conclusion ... 213

(13)

General conclusion ... 217

References ... 221

Annexes ... 223

Soft X-ray absorption (XAS, XMCD) principle ... 223

Principle ... 223

Detection modes ... 226

CROMAG End-station of DEIMOS beamline ... 227

Experiments ... 228

Polarized small-angle neutron scattering: principles ... 229

Annexes of Chapter I ... 234

Annexes of Chapter II ... 237

Annexes of Chapter III ... 243

SAXS details ... 243

TEM of CoO nanoparticles ... 243

Structural analysis of C2, CS2 and CS2 reheated ... 244

Fourier transform infrared ... 246

Granulometry... 246

Details on Stoner Wohlfarth fit of HC = f(T) of CS, CSSA, CSSB and CSSC ... 248

Ligand field multiplet (LFM) calculation ... 249

Annexes of Chapter V: Preliminary studies on the thermal decomposition of Ni based organo-metallic precursors ... 250

Synthesis of NiO nanoparticles in the literature ... 250

Chemical composition and ligands configuration of NiSt2 ... 251

Thermal stability of NiSt2 ... 252

Preliminary study on the growth of NiO nanoparticles from the decomposition of NiSt2 ... 252

Core@shell(@shell) nanoparticles synthesized from the thermal decomposition of NiSt2 ... 254

Fe3-dO4@NiO nanoparticles ... 254

Fe3-dO4@CoO@NiO nanoparticles ... 256

Discussion ... 257

Fe3-dO4@CoO@NiO nanoparticles synthesized with different concentration of NiSt2 ... 258

Determination of a new Ni precursor to synthesize NiO nanoparticles ... 259

(14)

Concentration effect ... 264

Spherical Fe3-dO4@NiO nanoparticles ... 266

FT-IR spectroscopy ... 266

XAS, XMCD spectroscopy... 267

Summary... 273

Experimental section ... 274

Nickel precursors ... 274

Iron and cobalt (II) stearate ... 274

Iron oxide core nanoparticles ... 274

Fe3-dO4@CoO core@shell nanoparticles ... 274

Fe3-dO4@NiO core@shell nanoparticles ... 275

Fe3-dO4@CoO@NiO spherical core@shell@shell nanoparticles (CSSNi4) ... 275

Fe3-dO4@CoO@NiO cubic core@shell@shell nanoparticles (CSSNi2A and CSSNi2C) ... 275

Fourier transform infra-red spectroscopy ... 276

Granulometric measurements ... 276

X-ray magnetic circular dichroism... 276

Magnetometry ... 276

Themogravimetry ... 276

References of annexes ... 277

Aknowledgements ... 283

(15)

(16)

General introduction

Owing to their size and shape dependent properties, magnetic nanoparticles have gained tremendous interest since the past two decades. Size reduction down to the nanoscale may disturbs the magnetic order and produce unblocked magnetic moment at room temperature. Indeed, the magnetic stability ordering versus temperature (kBT) depends on the volume of the nanoparticles (V) but also on their magnetic anisotropy constant (K). Thus, for small size, KV < kBT the magnetic ordering vanishes. Such property has found interest in biomedicine application. However, some other application such as data storage requires to get blocked magnetic moment over room temperature i.e. to get KV > kBT. In order to increase the magnetic stability of small nanoparticles against temperature, it is possible to tune their magnetic anisotropy constant by tuning their chemical composition. In this thesis, we are interesting to use iron oxide based magnetic nanoparticles which are made of a costless, abundant and natural material contrary to other commonly used rare earth based magnetic nanoparticles. However, small iron oxide magnetic nanoparticles of size of 10 ± 3 nm do not display magnetic ordering at room temperature. To increase their effective magnetic anisotropy constant, it is possible to synthesize bi-magnetic nanoparticles according to a core@shell model. Two different type of bi-magnetic coupling were investigated in this thesis: the exchange-bias coupling and the hard-soft coupling. The first one occurs between a ferrimagnetic phase and an antiferromagnetic phase and is further described in the general introduction. The second consists to benefit of a sufficient anisotropy difference between a hard and a soft magnetic phase and is described later.

Besides the requirements needed to generate strong magnetic couplings (exchange coupling or hard-soft coupling), the synthesis of bi-magnetic nanoparticles requires the two magnetic phases to display good epitaxial relationship i.e. crystallisation in similar space groups with close cell parameters. Three different compounds were selected: Fe3O4, CoFe2O4, CoO and NiO.

Previous studies on Fe3-dO4@CoO nanoparticles has showed that the growth of a CoO shell on the iron oxide core allows to increase the magnetic properties of iron oxide-based nanoparticles. This increase was attributed to a strong exchange magnetic coupling between the ferrimagnetic iron oxide core and the antiferromagnetic CoO shell which was favour thanks to the possible presence of interfacial diffusion. In this thesis, we have thus firstly investigated the diffusion mecanisms involved in bi-magnetic core@shell nanoparticles based on an iron oxide core.

Then, as the Fe3-dO4@CoO nanoparticles do not show a magnetic ordering at room temperature, the Fe3-dO4@CoO interface has been doubled in attempt to synthesize Fe3-dO4@CoO@Fe3-dO4 nanoparticles. It is expected here that the two ferrimagnetic/antiferromagnetic interfaces would increase the magnetic anisotropy constant further. To get rid of diffusion processes that may occur at both interfaces, Fe3-dO4@CoFe2O4@Fe3-dO4 nanoparticles were synthesized and their magnetic properties were compared to the previous Fe3-dO4@CoO@Fe3-dO4 nanoparticles. Finally, in order to benefit of the high magnetic ordering versus temperature of NiO, Fe3-dO4@CoO@NiO nanoparticles were synthesized.

(17)

The exponential development of devices for applications dealing with communications (data and motion) was allowed by the use of permanent magnets that are mostly composed of rare earth materials. In 2018, the world production of rare earth reached 170 000 tons, where China produced 71 % of it. The extraction of rare earth components requires to dig mines which adulterate and pollute the subsoil. Moreover, after extraction, their purification rejects a high volume of acids, bases, solvents, heavy metal and even radioactive materials. The purification processes also requires around 200 cubic meters of water which is at the end of the process, full of pollutants and often simply rejected. On top of that, the production of rare earth components emits a large amount of carbon dioxide which have to be decreased to limit the global warming according to the Paris agreements. Thus an alternative to produce new permanent magnets which are rare earths free has to be found. Owing to their unique structure, magnetic properties and low cost, spinel ferrites have already gained tremendous interest in different applications such as microwave,1_biomedical,2,3,3,4_sensors,5,6 high-frequency components,7_{supercapacitors}8,9_{and photocatalytic activity.}10,11_{Ferrites crystallize in the} Fd-3m cubic face centered AB2O4 structure where the 32 O2- anions form the cubic close-packed lattice, defining 64 tetrahedral (Td) sites and 32 octahedral (Oh) sites. However, only 8 Td sites and 16 Oh sites are occupied by cations. In a normal or direct spinel structure, the divalent transition metal cations (M2+_{) are located in Td sites while the trivalent Fe cations are in the Oh sites. The general formula is} thus written (!"#)[$%_"&#]'_*. At the opposite, in an inverse spinel structure the M2+_{share the Oh sites} with half of the Fe3+_{while Td sites are occupied by the rest of the Fe}3+_{cations according to} ($%&#_)[!"#_$%&#_]'

*. Ferrites may also display an intermediate structure as

(!+,-"# $%-&#)[!-"#$%",-&# ]'* where x is the inversion parameter. It is called partially inverse structure.

In the spinel structure, the magnetic moments supported by cations in Oh sites are parallel coupled one to the other through superexchange interactions via the oxygen anions. Such interactions is favored by the direct overlap of the d orbitals of the metal cations with the anion. The magnetic moment of cations in Td sites are opposed in sign to the magnetic moment of cations in Oh sites. Thus spinel ferrite are ferrimagnet. Therefore, the net magnetization is governed by the type of cations and their distribution in the crystallographic sites.

The most common spinel ferrites are composed of M2+_{= Mn, Fe, Co, Ni, Cu and Zn. They are} all in an inverse spinel structure except for ZnFe2O4 which displays a direct spinel structure. Their main characteristics are presented in

(18)

Table 1. According to this, they all present advantages and disadvantages:

Except ZnFe2O4, all Ferrites display very high Curie temperatures (TC) which corresponds to the transition temperature above which the ferrimagnetic order vanishes. Therefore, ferrites display permanent magnetization far above the room temperature. Due to a 4.6 µB magnetic moment, MnFe2O4 has the highest saturation magnetization (MS) of 111 emu/g. MS corresponds to the highest magnetic moment reached for a material submitted to an external induction due to the orientation of the spins along this induction. High MS are suitable for biomedical12–14 or permanent magnet applications.15,16_{In contrast, Zn and Cu ferrite are not interesting from this point of view. CoFe}

2O4 displays the highest magnetic anisotropy constant (K) which is a critical parameter for permanent magnets. However its cytotoxicity avoids biomedical applications and in this way, its use should be limited for every days devices. Although NiFe2O4 and Fe3O4 display the highest TC (858 K), Fe3O4 is usually preferred because of its higher magnetization saturation (MS) which is suitable for biomedical and permanent magnet applications. Furthermore, owing to its natural abundance, iron oxide is very cheap and can be produced with eco-friendly pathways. Iron oxide for advanced and day-life applications should be spread out.

(19)

Table 1. Structural and magnetic characteristics of principal spinel ferrites. a, TC, TN and µB stands for lattice parameters, Curie

temperature, Néel temperature and Bohr magneton, respectively.

a (Å) TC (K) Theoretical µB (0 K) Calculated µB(0 K) Experimental µB (0 K) MS bulk (0 K) (emu/g) K bulk (kJ/m3₎ MnFe2O4 8.513 585 5 4.6 4.6 111 -4 Fe3O4 8.396 858 4 4.1 4.1 98 20 CoFe2O4 8.392 790 3 3.7 3.7 94 220 NiFe2O4 8.337 858 2 2.2 2.2 56 -6.7 CuFe2O4 5.844 720 1 1.3 1.3 29 -6 ZnFe2O4 8.46 9 (TN) 0 0 0 - 0

Spinel ferrites crystallize in the MFe2O4 structure with trivalent Fe cations and M a divalent transition

metal. If the divalent metal is in the Td sites, the spinel structure is called direct while if the divalent metal is in an Oh site, the spinel structure is called invert. A mixed structure is characterized by an inversion parameter which has a high impact on the magnetic properties of the spinel. Moreover, the nature of the metal also modify the magnetic properties of the resulting spinel ferrite material.

Fe

3-d

O

4

structure

Crystallographic structure

The iron oxide (Fe3-dO4) spinel structure consists of iron cations with several oxidation states (Fe3+ and Fe2+_{) where non-oxidized Fe}

3O4 material is called magnetite with d=0 while the fully oxidized state with d=1, is called maghemite and displays the g-Fe2O3 general chemical formula. Magnetite crystallizes in the inverse spinel structure defined previously. Its’ primitive cell, shown in Figure 1, is composed of 56 atoms with 32 oxygen atoms with 8 Fe3+_{in Td sites, 8 Fe}3+_{and 8 Fe}2+_{in Oh sites.}17

(20)

The developed formula for magnetite is (Fe3+₎

Td[Fe3+Fe2+]Oh(O2-)Td while maghemite, which can be considered as a Fe2+_{deficient magnetite, displays no Fe}2+_{cations leading to the apparition of vacancies} denoted □. Contrary to the magnetite, three different structures of maghemite have been reported in the literature according to the disposition of the vacancies:

- within the same Fd-3m space group as magnetite and without any cell deformation, with a general formula ($%&#)_./0$%_1/&&#□_+/&4

56('",)*. This structure is favored by a similar

occupation rate of 5/619_{and is the major compound formed after the oxidation of magnetite.} - In the P4132 space group without any cell deformation, with a general formula

($%7&#)./0$%*/&&#□7/&$%+"&#4₅₆('",)&" favored by a partial order on the Oh sites.19,20 - In the P43212space group where the cubic cell becomes tetragonal.19,21

Magnetite is an abundant material which crystallizes in the inverse spinel structure. This material easily oxidize upon exposure to air to produce maghemite, leaving room to the apparition of vacancies instead of the originals Fe2+_.

Magnetic structure

Fe26_{displays a [Ar] 3d}6_4s2 _{electronic structure with 8 electrons in the external electronic layer. Hence,} Fe2+_{and Fe}3+_{are featured by 6 and 5 electrons in their external electronic layer represented in Figure} 2. According to this, Fe2+_{and Fe}3+_{cations support respectively 4 and 5 µ}

B of electronic moment.

Figure 2. Scheme of the electronic configuration of Fe2+_{and Fe}3+_{in the magnetite.}

As in the inverse spinel structure, the spins present in Td sites are opposed in sign to the ones in Oh sites. Fe2+_{and Fe}3+_{in Oh sites are antiparallel coupled through double exchange interactions. Hence,} in the magnetite structure, the magnetic moments of Fe3+_{cations cancelled each other’s. The net} magnetic moment of 4 µB is only given by the Fe2+. Thus, magnetite displays a ferrimagnetic (F(i)M) magnetic behavior.

The same ferrimagnetic behavior is observed in the ($%&#)_./0$%_1/&&#□_+/&4

56('",)* maghemite where

the magnetic moment of one Fe3+_{in Td sites is opposed to five third of the magnetic moment} supported by Fe3+_{in Oh sites, leading to a net magnetic moment of 3.33 µ}

B for the maghemite. Due to the invert spinel structure, magnetite displays a ferrimagnetic behavior leading to a net magnetic moment of 4 µB. While in maghemite the net magnetic moment decreases to 3.33 µB with the

(21)

Size reduction to the nanoscale

In order to be incorporated into devices such as computers, smartphones and so on, the size of iron oxide based magnet has to be very tiny. Size reduction to the nanoscale allows to open new frontiers. At the nanoscale, the effect of oxidation on the magnetite structure is spontaneous and has a higher impact than in the bulk form.14,22–24_{Indeed, it has been proved that below a size of 8 nm, an iron oxide} nanoparticle is only composed of maghemite. For sizes higher than 12 nm, it displays a magnetite structure with a maghemite shell according to a core@shell structure with a gradient of maghemite composition from the surface. For intermediate sizes, the composition of the nanoparticle is a mixture of both magnetite and maghemite, according to a general structure of Fe3-dO4.25 Such behavior modifies the magnetic properties of the nanoparticles.26

Magnetic domains size

Size reduction down to the nanoscale increases the surface over volume ratio which has the disadvantage to decrease the magnetic stability for increasing temperatures.

Indeed, in the bulk state, a magnetic material is composed of several magnetic domains called Weiss domains, which minimize the magnetostatic energy and are delimited by the Bloch walls. Inside each magnetic domain, there is a single magnetic orientation while the domains have different magnetic orientations between them. This magnetic structure leads to the absence of a spontaneous magnetization. However, under a critical radius, rC, which varies according to the chemical structure of the material, the material shows a single magnetic domain with a spontaneous magnetization.

89 =9√<>_µ ?!@"

with A the exchange constant, K the magnetic anisotropy constant, µ0 the magnetic permeability and MS the saturation magnetization.

If the size is even more reduced under a critical radius r0, the spontaneous magnetization at room temperature disappears, living room to the superparamagnetic state where the overall magnetic moment of the particle fluctuates over short times.

Figure 3. a) Magnetic behavior as a function of the size of the nanoparticles.27_{b) Maximum sizes of nanoparticles for single}

magnetic domains (rc) and superparamagnetic (rO) behavior according the chemical composition.28

In the bulk form, a material has no spontaneous magnetization due to the presence of several Weiss domain. Decreasing the size under a critical rc radius lead to the apparition of a single magnetic domain.

If the size is decreased further under r0, the spontaneous magnetization disappears, and the magnetic

(22)

Superparamagnetism

According to the Stoner-Wohlfarth model, size reduction below a critical volume (V) depending on r0, lead to the overall decrease of Ea defined as Ea = KV sin² θ with K the magnetic anisotropy constant of the material and θ the angle between the magnetization direction and the easy axis. Thus, if the thermal energy 25kBT is higher than Ea, the magnetic moment of the nanoparticles will easily shift from the spin up to the spin down configuration over time.29_{This regime is called superparamagnetism and} was predicted by Frenkel and Dorfman.30_{Such properties has found applications in biomedicine for} MRI contrast agents as an example31_{while they are not suited for others such as data storage where} blocked magnetic moment above room temperature are required. For nanoparticles, the transition temperature between the blocked magnetic state and the superparamagnetic state is called blocking temperature (TB) defined as

AB = >C

DEBFG HI_IJ_?KL

where the KV product corresponds to the magnetic anisotropy energy of the material, kB is the Boltzman constant and I_J and I_? are the time of measurements and the reversal attempt time (usually in the range of 10-9_{to 10}-12_{s) respectively. Hence, it shows that the superparamagnetic state is not} only proper to the material but also depends on the magnetic measurement technique where I_J is in the order of 100 s for SQUID measurements in DC mode and in the range 10-7_{to 10}-10_{s for Mössbauer} measurements. Where SQUID refers to super quantum interference device and is a magnetometry measurement technique. For SQUID measurements, the typical times used are I_J = 100 s and I_? = 10-9_{s, thus the equation becomes}

KV = 25kBTB

When the nanoparticles display a size distribution, it leads to a large volume (V) of distribution and thus the determination of TB is often challenging. Thus, in the literature, the Tmax is better compared than TB. Tmax is ascribed to be the maximum of the zero field cooled curve (ZFC) from magnetization versus temperature measurements. The width of this curve depending on the size distribution of the nanoparticles, Tmax corresponds to a distribution of blocking temperatures.

Figure 4. Anisotropy energy profile as a function of the angle between the magnetic moment and the easy axis for a single domain NP with pure uniaxial symmetry.32

(23)

More specifically, iron oxide displays superparamagnetic properties at room temperature for size smaller than 30 nm.33_{Such properties are evidenced by recording magnetization as a function of the} temperature (Figure 5) which also shows that the size reduction of the nanoparticles results in the decrease of Tmax.

Figure 5. a) Magnetization versus temperature curves for different size of iron oxide nanoparticles. b) Tmax as a function of the

size of iron oxide nanoparticles. [Baaziz reproducible tuning]

If a material is in a superparamagnetic state, the measure of its magnetization curve subjected to a reversing applied magnetic field, evidences a single and closed curve called hysteresis (Figure 6a) i.e. which does not have any coercive field.

Indeed, a hysteresis is characterized by three factors:

- A saturation magnetization (MS) (defined in the general introduction)

- A remanent magnetization (MR) which is the natural magnetization of the material without any applied magnetic field,

- And a coercive field (HC) corresponding to the reverse magnetic field needed to cancel the magnetization of the material.

According to this, a nanoparticle with a blocked magnetic domain shows an open hysteresis with a coercive field which at T = 0 K can be calculated according to the Stoner-Wohlfarth model:34

O9 ≈_µ>Q ?!@

Showing that this coercive field is proportional to the anisotropy constant of the material and to its saturation magnetization. According to this observation, a material featured by a low anisotropy constant and a high saturation magnetization, also called soft magnetic material, display a hysteresis which has a small coercive field. While a hard magnetic material featured by a high anisotropy constant and a low saturation magnetization displays a hysteresis curve that has a large coercive field (Figure 6b).

(24)

Figure 6. a) Magnetization curve of a superparamagnetic material. b) Magnetization curve of a hard and soft blocked material with the characteristics features of a hysteresis curves (MS, MR and HC).

Superparamagnetic properties are the results of a higher thermal energy than the intrinsic magnetic anisotropy energy of the considered material. The transition temperature between the blocked magnetic state and the superparamagnetic state of a nanoparticle is called blocking temperature. However, the superparamagnetic state is time-dependent and the determination of TB depends also on

the time of measurements of the analysis technic used to probe the magnetic properties of the material. In M(T) curves, Tmax is generally better used than TBas it is easier extracted. The close hysteresis in a

M(H) curve of a material evidences the presence of superparamagnetic properties.

Dipolar interactions

It is important to note that in this manuscript, the magnetic properties of the nanoparticles are recorded on powders (unless otherwise stated). Hence nanoparticles supporting a net magnetic moment are influenced by the presence of dipolar interactions which depend on the distance between the nanoparticles (d) and on their magnetic moment (µ). For spheres of radius r, the dipolar interaction energy depends thus on the saturation magnetization of the nanoparticles and on the separating edge-to-edge distance (s), which according to the 2-dipole approximation model gives:35

R/= µ "

S&=(!@× 4 3⁄ W8 &₎"

(28 + Z)&

It shows that dipolar interactions can be avoided by increasing sufficiently the distance between the nanoparticles, i.e. by diluting the nanoparticles in a matrix such as eicosane17_{or in a solvent or by} carefully assemble isolated nanoparticles on a substrate.36

Dipolar interactions correspond to the magnetostatic interaction between the magnetic moment of each nanoparticle. Thus according to interactions in a triangular lattices which is a more reasonable model than the 2-dipole approximation for nanoparticles in the powder state,35,37_{it results that:}

R/= 2.8 × 10,_µ "

S&

In consequence, dipolar interactions modify significantly the magnetic properties compared to isolated nanoparticles. Dipolar interactions results in the enhancement of the blocking temperature.38,39_They are also expected to decrease HC and the MR/MS ratio because of the collective properties which favor the easier reversal of magnetization.38,40

(25)

Although dipolar interactions markedly modify the magnetic properties of nanoparticles, it is difficult to rationalize their effect. Indeed the precise control of the spatial arrangement of nanoparticles (interparticle distance and the dimensionality of their assembly) over large areas is very challenging.

Magnetic nanoparticles can be subjected to dipolar interactions whose strength depends on the distance between the nanoparticles. Such interaction may affect the magnetic properties of the nanoparticles assembly. Thus, a precise control on the distance between the nanoparticles allow to better understand the magnetic properties of a unique nanoparticles or of a packed arrays of nanoparticles. Nevertheless, the control of the distance between nanoparticles over large area is often challenging.

Small iron oxide nanoparticles

For permanent magnet applications in data storage, very small iron oxide nanoparticles should be used in order to store a high density of information in the smallest space as possible. However, it was shown that for size smaller than 30 nm, these nanoparticles are superparamagnetic at room temperature. Thus, an alternative has to be found in order to increase the magnetic anisotropy energy of such nanoparticles.

Exchange-bias coupling in nanoparticles

Definition and characteristics

The interfacial exchange anisotropy mostly known as exchange-bias coupling has been discovered by Meiklejohn and Bean in 1956.41_{They studied the magnetic properties of ferrimagnetic (FiM) cobalt} nanoparticles surrounded by an antiferromagnetic (AFM) cobalt oxide shell. They noticed that for T > TN, where TN is the Néel temperature of CoO, the nanoparticles displayed typical properties of Co nanoparticles. In contrast, for T < TN, FiM-AFM spins interactions modify the magnetic properties of the Co@CoO nanoparticles.

Indeed, ferro(i)magnetic F(i)M nanoparticles generally evidence a uniaxial anisotropy with a nice sinusoidal applied magnetic field angular (φ) dependence behavior of their torque magnetic curve (Figure 7 a, b, c). When they are in contact with an AFM, for T > TN (Figure 7a), the magnetic torque curve evidences an increase in amplitude, and for T < TN, the torque curve displays a different number of stable positions (magnetic torque = 0) with variations of the amplitude.42,43

(26)

Figure 7. a, b, c) Magnetic torque as a function of j angle where φ is the angle between magnetization and applied magnetic field, d, e, f) magnetic moment versus an applied magnetic field (M(H)) curves of a, d) a ferrimagnet below TC, a ferrimagnetic

in contact with an antiferromagnet at T<TN b, e) after ZFC, c, f) after FC. Adapted from ref.42,43

Moreover, considering a ferro(i)magnet and an antiferromagnet with TC>TN as it is generally the case, a M(H) hysteresis cycle of a ferro(i)magnet in direct contact with an antiferromagnet recorded at T<TN after zero field cooling (ZFC) evidence an increase of the coercive field (HC) due to the pinning of the ferro(i)magnetic spins by the antiferromagnet (Figure 7 d, e). In contrast, after field cooling, the hysteresis curve is shifted to negative magnetic field which corresponds to the so-called exchange-field (HE) and is typical of exchange-bias coupling (Figure 7 f).

Such a phenomenon is allowed by the coupling of interfacial AFM spins with the interfacial F(i)M spins during the field cooling procedure. The F(i)M interfacial spins are pinned by the AFM spins if KAFMVAFM > KF(i)MVF(i)M, the AFM exerts a torque on the ferro(i)magnet resulting in the apparition of HE. However, if KAFMVAFM < KF(i)MVF(i)M, the AFM does not exert a sufficient torque on the ferro(i)magnet to pin the interfacial spins resulting in the solely increase of HC without any HE. A second condition to allow the presence of an exchange-bias interaction is that KAFMVAFM has to be superior to the interfacial coupling energy Jint defined as `_abc = (O_d!_@e) 6⁄ , where D is the size of the FiM core.42

In order to universally compare the strength of exchange-couplings between different materials, it is possible to calculate the coupling energy E=HEMSVF(i)M with MS the saturation magnetization and V the volume. It becomes E=HEMSdF(i)M/6 in the case of spherical core-shell nanoparticles.42 However, this is not a very accurate consideration as small variation of the initial spherical shape model has a strong influence on the physical properties. Furthermore, in this manuscript we are more interested in the direct comparison of the exchange-bias strength. Thus we will compare the values of HE which are directly proportional to the interface exchange anisotropy.44

(27)

Figure 8. Scheme of the spins configuration during the recording of the magnetic moment versus an applied magnetic field after field cooling at a FM/AFM interface a) with KAFMVAFM>KFMVFM, b) with KAFMVAFM<KFMVFM. Adapted from ref.42

Figure 8a presents the spin configuration at the FM/AFM interface during the recording of the M(H) hysteresis cycle after FC for a ferromagnetic interfacial exchange coupling. For KAFMVAFM > KFMVFM, it evidences that:

- 1/ The FM and AFM spins display a parallel configuration.

- 2/ The spins of the FM phase follow the applied magnetic field as it is reversed with some difficulties due to the torque exerted by the AFM.

- 3/ The FM spins display an antiparallel configuration compared to the AFM spins.

- 4/ The FM spins easier follow the reversing of the applied magnetic field than in (1) thanks to the help of the interfacial spins of the AFM which are this time oriented in the same direction as the applied magnetic field (1). Thus the spins configuration get back to their initial and lower magnetic configuration.

The exchange-bias is usually characterized by a negative shift of the hysteresis toward the applied magnetic field axis which allows to reach the saturation magnetization with lower applied magnetic field. However, a positive shift was also attributed to an antiferromagnetic interfacial coupling.45_At the opposite, for KAFMVAFM < KFMVFM (Figure 8b), after FC,

- A, C/ the spins are oriented in the same directions as in the previous case.

- B, D/ Then the spins in the FM phase follow the applied magnetic field reverse and exert a torque on the AFM spins’ which will follow the field reversing due to its’ lower magnetic anisotropy.

In this case, no exchange field is observed as the FM and AFM phases rotate coherently with the applied magnetic field.

The exchange-bias properties is the result of the spin pinning of ferromagnetic or ferrimagnetic phase by an antiferromagnetic counterpart. Such magnetic exchange coupling is observed if KAFMVAFM >

KFiMVFiM and if KAFMVAFM > Jint. It results in the increase of the overall magnetic anisotropy of the material

and allows to increase the TB of the ferrimagnet. The exchange-bias properties is characterized by a

shift of the hysteresis on the applied magnetic field axis in M(H) curves due to a resistance of the FiM spins to follow magnetic field reversal.

(28)

Limitations of the exchange-bias coupling

Effect of temperature

The exchange-bias coupling is the result of the interfacial spin pinning of the F(i)M phase by the AFM one. Hence when T>TN, the AFM phase loses its’ magnetic stability and displays paramagnetic properties. Thus the AFM is no longer able to drive the interfacial pinning and the exchange-bias coupling becomes absent.

According to this consideration, from low temperature, HE and HC are expected to decrease when T approaches TN. It is observed that HE vanishes before reaching TN while HC can be present up to T=TN.46,47 The temperature at which HE vanishes is also called TB in the literature and needs to be differentiate to the TB representing the limit between an overall blocked magnetic state and a superparamagnetic state. Nevertheless, the disappearance of HE before the one of HC can be attributed to the eventual presence of small AFM crystallite which are in a superparamagnetic state phase for T<TN.48

Volume effect

The exchange-bias coupling depends on the volume of the AFM and F(i)M materials as it is an interface effect. Indeed, considering KAFM>KF(i)M, there exists a maximum and a minimum volume of the F(i)M where HE vanishes. The maximum volume of F(i)M is reached for volumes that are equals to the F(i)M domain wall volumes. And the minimal volume may be reached if the F(i)M phase displays a discontinuity.49_{In between, H}

E is proportional to 1/VF(i)M and is enhanced for small volume of F(i)M such as HC which follows the same evolution as HE.50

Moreover, there also exists a minimal volume of AFM for which KAFMVAFM is inferior to the interfacial energy coupling JFM-AFM resulting in the vanishing of HE. This minimal volume is defined as VAFM=J FM-AFM/KAFM.51,52

Training effect

Training effects have been observed in thin films53–55_{and in nanoparticles}56,57_{and correspond to a} decrease of HE and HC for a successive repetition of M(H) loops. This is attributed to a deviation of the interfacial spins from their equilibrium configurations due to a relaxation phenomenon of the interfacial frozen spins along the field cooling direction.50_{Larger training effects are generally observed} for thin AFM thicknesses.53,54

Field cooling effect

Different behavior of HE towards the increase of the cooling field were reported.

In core@shell cobalt ferrite nanoparticles, it is possible that HE increases up to 1.5 kOe with the increase of the field cooling to 5.0 kOe and decreases for higher fields.58_{The increase of H}

E arises from an increase of the number of FiM spins aligned with the applied magnetic field. While the decrease is attributed to a decorrelation of the interfacial AFM spins that tend to align with the applied magnetic field.

Moreover, in FeF2-Fe bilayers, Nogués and al. have reported a negative exchange field for small cooling fields while for cooling fields higher than 20 kOe, they reported a positive exchange-bias which arise from an AFM interface coupling.45_{The same behavior has been reported for iron and manganese oxide} based core@shell nanoparticles.46_{Thus field cooling has an effect on the measured H}

E and it is necessary to record the HE in the exact same condition.

(29)

Asymmetry

Hysteresis curves can display some asymmetry related to the presence of differences in the magnetic reversal process between the two branches.50_{Different mechanisms can occur in such conditions:}

- A uniform spin rotation appears for the reversal in the decreasing field branch while in the increasing field branch, the spins’ reversal are subjected to nucleation and propagation of domains wall

- It can also be produced by a competition between the ferromagnet and the interfacial FM-AFM interfacial anisotropies.

Spins orientation

Exchange-bias results from the parallel coupling of interfacial F(i)M-AFM spins. However, it is also possible that the spins display a different orientation with a perpendicular direction of the AFM toward the F(i)M.59,60_{It results from the rotation of the AFM or F(i)M during the field cooling procedure and} was predicted to occur for systems with a low F(i)M magnetic anisotropy.61_{It induces a minimal value} of the interfacial energy for perfectly compensated surface and can thus increase the exchange bias.61

Vertical shift

Hysteresis loop may also show a shift along the magnetization axis called vertical shift (HV). It has been proved that the vertical shift is influenced by the field cooling where it is negative for low field cooling, positive for high field cooling and arises from uncompensated spins momentum in the AFM.62_Hence the vertical shift is proportional to the number of uncompensated spins.50

Interface disorder Roughness

The increase of interfacial roughness induces some magnetic interfacial defects that generally decrease the exchange field due to the presence of magnetic frustration as shown by Figure 9.

Figure 9. Schematic representation of the interfacial spin configurations with magnetic defects. Adapted from ref.63

However, it has also been reported that some systems are not influenced by the increase of interfacial roughness.64_{Or that the exchange bias coupling increased with the increase of the interfacial} roughness due to the pinning of the propagating domain walls in the ferromagnet.65

(30)

Diffusion

Interfacial atomic graded composed interfaces,66_{mixed interfaces}67_{and diffusion,}68_{have shown the} enhancement of coercive field while the exchange field and so the exchange coupling was reduced. The role of the interfacial diffusion is further studied in Chapter II.

The exchange-bias coupling finds numerous parameters where their control allows to finely tune the resulting properties in order to use it for different magnetic applications.

Applications

The exchange-bias property finds applications for read head,69_{giant magneto-resistance}69_and MRAM.70_{Moreover, in the past years, this property has also found interest in permanent magnets}71 and data storage72_{thanks to the enhancement of the coercive field and blocking temperature.}49_Owing to the improvement of the digital technology and the exponential increase of the numerical tools used by the world population, the need to extend the data storage capability is crucial. A solution could be the use of magnetic nanoparticles. However, most of them are based on rare-earth materials which on top of that display low energy product (BH)max.73 A high energy product in the hysteresis ensures the magnetic stability of the nanoparticles over time allowing to store the information for long time. Thus, the development of rare-earth free small nanoparticles with a stable magnetic moment above room temperature and a high energy product is of interest.

Exchange-biased iron oxide based magnetic nanoparticles

Iron oxide nanoparticles are particularly attractive due to their biocompatibility and abundance. They display a ferrimagnetic order which, coupled to an antiferromagnet, is expected to produce exchange bias coupling. Thus the magnetic properties or iron oxide based nanoparticles can be tuned with this exchange coupling property.

As the exchange-bias is an interfacial coupling effect, it is required to get an intimate contact between the iron oxide nanoparticles and the AFM material. According to this, it is possible to insert the F(i)M nanoparticles in an AFM matrix74_{or to synthesize FiM@AFM core@shell nanoparticles which have the} advantage to be more modular. Indeed, with core@shell nanoparticles, it is possible to tune the core size, the shell thickness, and the distances between the nanoparticles which all have a high impact on the magnetic properties.

In order to get a smooth FiM/AFM interface in iron oxide@AFM core@shell nanoparticles, the AFM phase needs to be cautiously selected. Indeed, to grow the shell on the core, a good epitaxial relationship is required: the AFM also has to crystallize according to a cubic structure with a cell parameter displaying the lowest lattice mismatch as possible. Moreover, to display exchange-bias property, the effective magnetic anisotropy energy of the AFM phase has to be so that KAFMVAFM>J FiM-AFM and that KAFMVAFM>KF(i)MVF(i)M. On top of that, the AFM should display a high TN as the blocking temperature in such system is generally equal to TN because above TN, the AFM order vanishes, and consequently the exchange bias.

According to the previous paragraphs, iron oxide based nanoparticles featuring exchange-bias properties are listed in Table 4.

(31)

Surface anisotropy

Surprisingly, it was found that simple FiM nanoparticles of g-Fe2O3,75–77 Fe3-dO425 and CoFe2O458 are also able to display an exchange field without the need to be in contact with an antiferromagnet. So does with iron oxide nanoparticles subjected to a 5 GPa pressure.78_{The origin of their exchange bias} property lies in the presence of disordered spins at their surface, resulting in spin canting effects. They are the result of a break of magnetic symmetry at the surface of the nanoparticles. Canted surface spins align then with the external field during the field cooling procedure creating a magnetic core@shell structure. The surface spins exert then a magnetic torque on the interfacial internals spins leading to a small exchange field. However, even if the presence of exchange-bias in these nanoparticles increase their coercive field, it does not have any significant impact on their blocking temperature which remains very low.

Exchange-bias coupling were observed in simple material nanoparticles that results from spin canting effect which generate a magnetic core@shell structure.

Phase oxidation

Most of the studied systems that display an exchange field are based on the spontaneous or forced oxidation of the native material as first reported for the Co@CoO nanoparticles studied by Meiklejohn and bean.41

Interestingly, FeO is an AFM material that spontaneously oxidized into FiM Fe3-dO4. According to the size of the native FeO nanoparticles, it is possible to form a stable FeO@Fe3-dO4 system. It has been reported that small FeO@Fe3-dO4 nanoparticles with a core size of 7 nm and a core@shell size of 14 nm, display low HC, HE and TB of 0.7 kOe, 0.5 kOe and 100 K respectively.79 The increase of the size to 16.2 nm of cubic FeO@Fe3-dO4 nanoparticles evidenced an increase of HC, HE and TB to 1.9 kOe, 0.6 kOe and 209 K respectively.80_{In the previously cited nanoparticles, it is possible that the cubic shape of the} nanoparticles has tuned the magnetic properties of the FeO@Fe3-dO4 nanoparticles compared to a spherical morphology of the same volume.81_{Bigger spherical FeO@Fe}

3-dO4 nanoparticles with a core size of 10 nm and a core@shell size of 23.2 nm show a further increase of HC and HE to 2.3 kOe and 1.6 kOe and of TB over 275 K.82 However, the last nanoparticles do not show a significant improvement of the magnetic stability with respect to temperature compared to pure Fe3-dO4 nanoparticles of 21 nm which display a TB of 248 K.25 As the TN of FeO is 198 K, it is not expected that FeO still pin the spins of the Fe3-dO4 FiM shell above this temperature. Thus the high TB of the two last mentioned nanoparticles is more related to volume effects than to an efficient exchange-bias coupling.

Fe is a ferromagnet (FM) which easily oxidized into Fe3-dO4. Thus Fe@Fe3-dO4 nanoparticles which consists of a FM@FiM material can be synthesized. Hence no exchange bias property are expected while some rather high HE in the order of the 1-6 kOe were measured for core@shell sizes of 11 to 13.8 nm.56,83_{The same behavior was observed in FiM@FiM Fe}

3O4@g-Fe2O3 nanoparticles. Indeed, as magnetite also spontaneously oxidized into maghemite, it has been reported that Fe3O4@g-Fe2O3 nanoparticles with a size of 12 nm evidence a low HE of 0.1 kOe with a low TB of 180 K.84 Moreover, it was reported that ultrasmall MnFe2O4@g-Fe2O3 and CoFe2O4@g-Fe2O3 with a size of 3.6 and 4.3 nm also display small HE of 0.09 and 0.12 kOe respectively.85 The intriguing presence of exchange-bias in such F(i)M@FiM nanoparticles was attributed to spin-glass-like phase which has a higher magnetic anisotropy energy than the rest of the FiM phase in the nanoparticles. The spin glass like phase orientates with the external field during the field cooling process and acts then as the pinning layer.56 Even if this property is very interesting in such system, the TB of such nanoparticles remains well below room temperature.

(32)

The natural oxidation of a magnetic material alters its chemical structure but also its magnetic structure. Hence it is possible that a native AFM material oxidized into a ferrimagnet, and generates an AFM/FiM interface resulting in the appearance of exchange-bias coupling.

Synthesis of iron oxide nanoparticles

Besides the oxidation of a magnetic phase, recent advances in synthesis techniques has allowed to go deeper by controlling the formation of a shell with different chemical compositions and with a good epitaxial relationship. The variety of combination between a hard AFM and a soft FiM phases has opened huge perspectives towards the enhancement of the magnetic properties of nanoparticles through exchange bias coupling. Seed-mediated growth methods were developed in order to synthesize iron oxide based core@shell nanoparticles that does not result from the oxidation and are grown at the surface of a ferrite spinel structure.

The seeds nanoparticles can be synthesized according to various techniques which are presented in Table 2. Each one has its own pro and cons. In this thesis, the nanoparticles have a size smaller than 20 nm with a high control of their size and size distribution. Furthermore, their morphology should be controlled as well and the synthesis should produce nanoparticles with a yield as high as possible in order to use them for potential applications. Thus the thermal decomposition appears to be a technique of choice.

Table 2. Summary of the different synthesis conditions for inorganic nanoparticles.17 Synthesis method Synthesis conditions T (°C) Reaction time Solvent type Size (nm) Size control Morphology control Yield

Coprecipitation Very simple 20-90 Very short aqueous < 20 Not well

controlled Medium High

Microemulsion Difficult 20-50 Short aqueous/

organic < 50 Narrow Good Low

Polyol Very simple > 180 Short organic < 10 Narrow Very good Mediu

m

Hydrothermal Simple > 200 Hours aqueous /

ethanol < 1000 Narrow Very good

Mediu m Thermal decomposition Difficult 200-350 Hours organic < 40 Highly

controlled Very good High

Thermal decomposition method

The thermal decomposition method is based on the decomposition of an organic metallic precursor in a solvent at high temperature, typically from 290 to 350 °C. Thus organic solvents with high boiling temperatures are considered. The principle of this synthesis is governed by the LaMer theory86 described in Figure 10 which was directly observed in solution thanks to in situ small-angle scattering.87 When the solution containing the metallic precursor is heated to high temperature, the precursor starts to decompose and to form monomers which are the smallest building units. The concentration of the monomers then increase gradually in solution up to a critical supersaturation concentration (SC) where the energy is high enough to overcome the energy barrier of the burst of nucleation. The monomers aggregate forming nucleus and their concentration in solution rapidly drops down to SC, where the homogeneous nucleation of the monomers stop and the growth step starts. This last is favored by the fact that the energy barrier of growth is much lower than the energy barrier of nucleation. As the nucleus were formed simultaneously and rapidly, they all display the same size which allow the nuclei to grow at the same time, thus giving rise to the narrow size distribution of the nanoparticles.88,89_{If the reaction time is continued during the growth step, the nanoparticles are} subjected to a new process called Ostwald-Ripening where the smallest nanoparticles are resolubilized in solution in order to grow on the biggest ones.

(33)

Figure 10. Schematic representation of the LaMer theory governing the synthesis of nanoparticles in solution.88

The thermal decomposition method allows to produce a large quantity of controlled size nanoparticles in a few hours. It consists in decomposing an organo metallic precursor in a high boiling point solvent with the eventual presence of a surfactant. This synthesis is ruled by the LaMer theory where the monomer concentration in solution allows to control the nucleation and growth step and de facto the structure of the resulting nanoparticles.

Seed-mediated growth synthesis by thermal decomposition

The synthesis conditions of the thermal decomposition may be modified by the presence of seeds in the reaction medium. According to ref,90_{the seed-mediated growth in solution occurs in three stages} (Figure 11):

1/ The heating of a solution containing seeds and metallic precursors generates monomers, leading to a first burst of nucleation. This heterogeneous nucleation allows to block the seed from a rapid growth and will later act as a monomer tank for the growth of the nanoparticles. The primary nucleation is then followed by a second burst of nucleation located at the surface of the seeds.

2/ An intraparticle ripening process occurs which is different from Ostwald ripening. There, the smallest particles synthesized from the first burst of nucleation dissolve. The size-dependent dissolution of the nanoparticles is proportional to 1/(exp(diameter)) according to the Gibbs-Thomson equation.91_{Their dissolution participates to a dynamic equilibrium of the monomer concentration} which is mainly governed by their diffusion and allows the growth of the remaining nanoparticles.

3/ Once the monomers are entirely consumed, the small nanoparticles dissolve at the expense of the larger ones, according to the Ostwald ripening process.

(34)

Figure 11. Schematic representation of a seed-mediated growth synthesis. From ref.90

In thermal decomposition, the seed-mediated growth synthesis lowers the energy barrier of nucleation and helps to decompose an organo-metallic precursor.

MnO, CoO and NiO wüstite phases as shells

Thus the thermal decomposition method allows to finely control the structure of the nanoparticles. However, in order to synthesize hybrid exchange-biased core@shell nanoparticles, the shell has to display good epitaxial relationship with the Fe3-dO4 core. In the literature, three components were mainly reported: MnO, CoO and NiO.

Figure 12. Representation of the primitive cell of a NaCl type structure in the case of MnO, CoO or NiO.

They all crystallize in the cubic Fm-3m space group according to the cubic centered face NaCl structure. The oxygens and metallic ions both form two networks that are nested one into the other. Hence, oxygens and metal ions are all in Oh sites. MnO, CoO and NiO are featured by cell parameters of 4.446 Å (JCPDS card n°04-005-4310), 4.2612 Å (JCPDS card n° 70-2856) and 4.1771 Å (JCPDS card n° 47-1049) respectively. As they crystallize in a similar space group as magnetite (Fd-3m) and their cells parameters match well with the 8.396 Å of magnetite nanoparticles, good epitaxial relationships between Fe3-dO4, MnO, CoO and NiO are expected.

(35)

Table 3. Main structural and magnetic characteristics of MnO, CoO and NiO. Cell parameter (Å) K (kJ/m3₎ _T N (K) MnO 4.4460 2.8 10-2 ₁₁₈ CoO 4.2612 5.0 102 ₅₂₅ NiO 4.1771 8.0 290

In the MnO, CoO and NiO structures, the oxygens and metal cations, which are first neighbors, are ferromagnetically coupled. While two atoms that are second neighbors are antiferromagnetically coupled through super-exchange interactions. Thus MnO, CoO and NiO are antiferromagnets.

Moreover, the magnetic anisotropy constants are 0.028, 500 and 8 kJ/m3_{for MnO,}92_CoO41_{and NiO}93 respectively.

Concerning magnetite and maghemite, Kmagnetite = 11-13 kJ/m394 and Kmaghemite = 5-15 kJ/m376,95 in the bulk form. Thus, KMnO << KNiO < Kmagnetite/maghemite << KCoO. And, TN of bulk MnO, NiO and CoO is 118, 525 and 290 K respectively while small iron oxide nanoparticles of 10 nm diameter display a TB of 150 K.25 Thus, TN (MnO) < TB (Fe3-dO4, 10 nm) < TN (CoO) << TN (NiO).

Wüstite MnO, CoO and NiO phases all crystallise in a similar space group as Fe3-dO4 with a good

matching of their cell parameters. This should allow to get good epitaxial relationship between the phases in core@shell nanoparticles based on Fe3-dO4 nanoparticles. Furthermore, thanks to their AFM

property, it is expected that the core@shell nanoparticles display exchange-bias properties in order to increase their magnetic order with respect to temperature.

Fe3-dO4@MnO nanoparticles

According to this, Fe3-dO4@MnO nanoparticles were synthesized.67,96 In such nanoparticles, the exchange field can be very low (0.07 kOe67_{) or very high (5.9 kOe}96_{) that we attribute to the structure} of the core@shell nanoparticles. However, in both cases the exchange-bias coupling does not have a real impact on the blocking temperature of the nanoparticles were the maximum reaches 70 K.96_Such consideration is attributed to the low magnetic anisotropy energies of MnO and magnetite and to the low TN of MnO. Indeed, for an efficient exchange-bias coupling, KAFMVAFM should be superior to KFiMVFiM and to Jint and, TB generally reaches a maximum which corresponds to TN. Thus, they do not allow to increase so far the magnetic anisotropy energy of such core@shell nanoparticles.

The low magnetic anisotropy of Fe3-dO4 and MnO and the low TN of MnO does not allow to increase the

magnetic properties of the native iron oxide nanoparticles. Fe3-dO4@NiO nanoparticles

Owing to the high TN of NiO, Fe3-dO4@NiO nanoparticles were expected to display high TB.67,97,98 It appears that the core@shell nanoparticles all display very low exchange field arising from weak exchange bias coupling. This is attributed to the anisotropy constant of magnetite and NiO that are very close and that the condition KAFMVAFM>JFiM-AFM is not respected. Although TN of NiO is much higher than room temperature, the TB of the native iron oxide nanoparticles was not increased by the core@shell structure.67,97

Due to similar magnetic anisotropy of NiO and Fe3-dO4, the KAFMVAFM > Jint condition is not satisfied,

(36)

Fe3-dO4@CoO nanoparticles

According to the very high magnetic anisotropy of CoO (500 kJ/m3_{) compared to magnetite (11-13} kJ/m3_{), Fe}

3-dO4@CoO nanoparticles evidenced the highest exchange field of the iron oxide based core@shell nanoparticles. Indeed, HE reaches 4.3 kOe for a core size of 9.6 nm and a 1.5 nm thick shell.52_{Moreover, due to a high exchange coupling effect, the H}

C of the native iron oxide nanoparticles has been drastically increased from 0.4 to 12.7 kOe for the best structure which consists in a core with a size of 8.2 nm and a shell thickness of 1.0 nm.52_{This coupling also allowed to increase T}

B from 150 K for the 10 nm iron oxide nanoparticles up to 290-300 K for the Fe3-dO4@CoO nanoparticles.47,67 Such improvements are attributed to the high pinning effect of the CoO shell with KAFMVAFM>>JFiM-AFM and also to a good crystallinity of the interface favored by the good epitaxial relationships.

Thanks to the high magnetic anisotropy and high TN of CoO compared to Fe3-dO4, the TB of Fe3-dO4@CoO

nanoparticles was successfully increased up to 290 K as for HC and HE which appears to reach very high

values.

AFM@FiM core@shell nanoparticles

Even if this thesis aims at studying the structure – magnetic properties relationship of FiM@AFM exchange-biased core@shell nanoparticles, exchange bias property were of course observed in inverted AFM@FiM ferrite based nanoparticles. Indeed, we have already discussed on the presence of exchange-bias properties in FeO@Fe3-dO4 nanoparticles (Table 4) which were synthesized thanks to the natural oxidation of the native AFM FeO nanoparticles. However, the main disadvantage of these nanoparticles is that the FeO phase is unstable and turns to the Fe3-dO4 when exposed to air.

Panagiotopoulos and al.99_{reported on the direct comparison of the magnetic properties of} g-Fe2O3@CoO (FiM@AFM) and CoO@g-Fe2O3 (AFM@FiM) synthesized by thermal decomposition. They showed that in both cases, the dependence of HE and HC with respect to temperature is similar, describing an exponential decay. Unfortunately, due to the size difference between the two systems and to different dipolar interactions in the sample, they were not able to give a clear comment on the TB properties.

In CoO@CoFe2O4 nanoparticles, no exchange field has been reported.100–102 Such observation has been attributed to the higher interfacial coupling energy (Jint) than the magnetic anisotropy energy of the antiferromagnetic core (KAFMVAFM) and to similar magnetic anisotropy energy between both counterparts (KAFMVAFM≃ KFiMVFiM). It was also observed that the TB of the CoO@CoFe2O4 nanoparticles was higher than room temperature. It is worth noting that the ZFC M(T) curve evidenced a kink at 300 K which was attributed to the losse of the magnetic stability of the AFM CoO phase (TN = 293 K). However, the absence of exchange bias coupling (no HE and TB > TN) was not concomitant with the absence of magnetic coupling within the core@shell nanoparticles. Indeed, CoO@CoFe2O4 nanoparticles evidenced very high HC in the order of 28 kOe at 5 K for a total diameter of 7 nm and a 2-3 nm thick shell.100_{The authors attributed such high H}

C and TB to the strong magnetic exchange coupling between the AFM core and the FiM shell. Such a result is supported by a study on size effects on CoO@CoFe2O4 nanoparticles.101 The core size was tuned between 2.6 and 6.0 nm while the shell thickness was increased from 1.3 to 2.7 nm. While no HE was measured whatever the core size of the nanoparticles and their shell thicknesses, HC decreased from 30.8 to 21.5 kOe (at 5 K), and TB increased from 167 to 388 K with the nanoparticle size. They also attributed this behavior to arise from a very strong magnetic exchange coupling between both phases, even for T>TN, and to “the formation of a highly crystalline magnetic phase, which improves the coupling at the interface”.

The gradual replacement of some Co atoms from the CoFe2O4 shell in the CoO@CoFe2O4 nanoparticles according to CoO@ZnxCo1-xFe2O4 nanoparticles with x = 0, 0.25, 0.5, 0.75 and 1.0 102 showed very high